unRAID to FreeNAS. Only one question left

Status
Not open for further replies.

tgggd86

Cadet
Joined
Dec 26, 2012
Messages
2
So I've been reading as much as I can, without physically using FreeNAS, but can't find a clear answer to my question.

Is the file system expandable? Some of the documentation I read suggests yes, while others say yes, but you will lose data.

For example lets say I have my Nas in RAID-Z5 with a storage pool size of 6TB (1x3TB, 1x2TB, 1x1TB), and want to add a forth drive that is 2TB to increase the pool size to 8TB. Is this possible without losing data?

Thanks in advance.
 

toddos

Contributor
Joined
Aug 18, 2012
Messages
178
First, read this.

Second, pools with mismatched drive sizes will use the smallest single drive size. So your 3TB + 2TB + 1TB setup in RAIDZ1 configuration will yield 2TB of usable space (2TB of the 3TB drive will be "wasted", as will 1TB of the 2TB drive; the third "missing" TB is used for parity).

Third, ZFS is not easily expandable. You can safely increase the size of the drives in your pool (in the previous example, replace the 2TB and 1TB drives with two 3TB drives and you'll get 6TB of usable space), but that's pretty much it. Adding a single drive at a time can't be done safely.
 

tgggd86

Cadet
Joined
Dec 26, 2012
Messages
2
Thanks toddos, I read that link and still had a bunch of questions that you clarified above.

Which leads me to what I was thinking; FreeNAS is definitly not for me unless I plan on spending insane ammounts of money every time I want to increase my storage pool.
 

louisk

Patron
Joined
Aug 10, 2011
Messages
441
First, read this.

Second, pools with mismatched drive sizes will use the smallest single drive size. So your 3TB + 2TB + 1TB setup in RAIDZ1 configuration will yield 2TB of usable space (2TB of the 3TB drive will be "wasted", as will 1TB of the 2TB drive; the third "missing" TB is used for parity).

Third, ZFS is not easily expandable. You can safely increase the size of the drives in your pool (in the previous example, replace the 2TB and 1TB drives with two 3TB drives and you'll get 6TB of usable space), but that's pretty much it. Adding a single drive at a time can't be done safely.

Not sure I quite follow here. You can additional vdevs to a pool easily. Add drives, and then follow the webby.
I agree about adding a single drive. No parity means disaster when it fails.
 

toddos

Contributor
Joined
Aug 18, 2012
Messages
178
Not sure I quite follow here. You can additional vdevs to a pool easily. Add drives, and then follow the webby.
I agree about adding a single drive. No parity means disaster when it fails.

Even if you add mirrored or parity vdevs to a pool, you're still adding an extra point of failure. If any vdev dies in your pool, you loose it all. So let's say you have two RAID-Z1 vdevs merged together in a pool, and one of the drives dies. While you're resilvering after replacing the drive, a second drive dies in that same vdev. You've now lost the ability to recover that vdev, and you've lost everything that was on the other vdev in the pool.

If you have redundancy in your vdevs, you're betting that catastrophe is limited and can be corrected before a second failure (or third failure, if you're using RAID-Z2). If you have no redundancy in your vdevs, you're running on borrowed time. A drive will die, eventually, and then you lose everything.
 

louisk

Patron
Joined
Aug 10, 2011
Messages
441
Best practice dictates you not have more than 8-10 spindles in a vdev. In order to get any significant performance or storage capacity, you need multiple vdevs. zfs was designed this way intentionally by people who understood the ways that data grows. You talk about losing a drive in a vdev compromising the reliability of the pool, but you have the same reliability one of those vdevs. The idea is that you place redundancy in the vdevs, not in the pool. You can choose from 100% in a mirror to somewhere around 20% with a 10 spindle raidz2. You also increase performance by adding additional vdevs to the pool (performance of a vdev is that of the slowest device in the vdev because ZFS writes in parallel to all drives in a vdev at the same time).
 

Brand

Moderator
Joined
May 27, 2011
Messages
142
For example lets say I have my Nas in RAID-Z5 with a storage pool size of 6TB (1x3TB, 1x2TB, 1x1TB), and want to add a forth drive that is 2TB to increase the pool size to 8TB. Is this possible without losing data?

You seem to be confused on what a RAID-Z is because you example is not using a RAID-Z5.
 
Status
Not open for further replies.
Top